A data set encompassing the needs of the patients regarding tablets or capsules was created through a questionnaire survey conducted from December 1, 2022, to December 7, 2022, via harmo, an electronic medication notebook smartphone app operated by harmo Co, Ltd (part 1 in Figures 1 and 2). The data set also included the PHR, such as the dispensing history of the respondent users and drug formulation information in the package insert.

The target population comprised 49,505 men and women aged 20 years or older who had been prescribed either tablets or capsules within 90 days from 2 weeks (November 16) before the survey distribution date (December 1). We sent notifications about the questionnaire to the target population, including a link to the survey details, through the harmo smartphone app. Users were then asked to provide consent and respond through the linked platform. The questionnaire included the following items: whether they found the tablets or capsules too large to swallow, the names of the medications that corresponded to the above question, and whether they had swallowing difficulties. Since larger numbers of questions are associated with lower response rates, it was necessary to set the lowest number of questions possible. Thus, we set simple questions to evaluate whether dysphagia was present without using objective measures such as the 10-item Eating Assessment Tool in this study [8].

For PHR, we obtained the following information from the data stored in the electronic medication notebook (part 2 in Figures 1 and 2) while maintaining anonymity: dispensing history and basic information (gender and age).

For medication formulation information, we obtained the following information from the package inserts (part 3 in Figures 1 and 2): long diameter, short diameter, and thickness (Figure 3). At this time, we distributed tablets and capsules of the same formulation. This was because a previous study conducted by Kabeya et al [4] demonstrated that the difficulty of swallowing tablets and capsules is almost identical. The diameters were calculated as follows: diameter = long diameter = short diameter for regular circular formulations; capsule diameter = short diameter = thickness for capsule formulations; and diameter = long diameter = short diameter = thickness for spherical formulations.

The primary outcome was the risk factors affecting the difficulty in swallowing tablets or capsules owing to their large size. The secondary outcomes were an appropriate index of the size of tablets or capsules difficult to swallow due to their large size, such as “long diameter + short diameter,” and a threshold for the size of tablets or capsules difficult to swallow due to their large size.

We identified the names of medications in the dispensing history of those medications that were reported to be difficult to swallow owing to their large size in the questionnaire. Tablets or capsules with identifiable “long diameter,” “short diameter,” and “thickness” from the package insert were defined as “tablets or capsules difficult to swallow due to their large size” (parts 4 and 5 in Figure 1).

Tablets or capsules that could be identified from the dispensing history as being taken by the same patient or at the same period as tablets or capsules difficult to swallow due to their large size were defined as “concomitant tablets or capsules” and formed the control group (parts 6 and 7 in Figure 1). In Japan, tablets and capsules are sometimes powdered to make them easier to ingest, for example, by crushing the tablets or removing the powder from the capsule. In this study, we handled tablets and capsules as powders if instructions related to crushing could be obtained from the dispensing comments in the dispensing history.

Based on the medication formulation information, 6 indices were set for each of the tablets or capsules that were difficult to swallow owing to their large size and concomitant tablets or capsules (used as controls; parts 8 and 9 in Figure 1), and the receiver operating characteristic (ROC) was calculated for each index. The index with the largest area under the receiver operating characteristic curve (AUC) was used as the best index of the size of tablets or capsules difficult to swallow because of their large size (part 10 in Figure 1).

From the generated ROCs, the point with the highest discriminative performance that maximized the Youden index was identified, and the optimal threshold for each index was calculated (part 11 in Figure 1) [9].

Multivariate logistic regression analysis was performed to identify risk factors for patients who tended to find swallowing difficulties due to the size of the tablets or capsules. Based on information on patient preferences regarding tablets or capsules, the objective variable was defined as the presence or absence of tablets or capsules that were difficult to swallow because of their large size (part 4 in Figure 2). The following explanatory variables were set from patient preferences regarding tablets or capsules, PHR, and formulation information (part 5 in Figure 2): gender, age, dysphagia, number of oral medications per day (including powdered medicine and liquid), number of tablets or capsules per day, number of tablets or capsules per timing, total size of tablets or capsules per day, taking powdered medicine, taking liquid, single-packaging, and taking large tablets or capsules (Note S1 in Multimedia Appendix 1).

Among these explanatory variables, “number of oral medications per day (including powdered medicine and liquid),” “number of tablets or capsules per day,” “tablets or capsules per timing,” and “total size of tablets or capsules per day” were assumed to have correlations, and therefore were selected from the clinical perspective that the correlation ratio with “taking large tablets or capsules” was the lowest and could be easily calculated by health care workers. For variables that were not binary, cutoff values were determined based on the results of stratified group comparisons and were classified as binary (part 6 in Figure 2). In addition, sensitivity analysis was conducted to evaluate whether different factors were extracted when patients were stratified based on their age (by dividing the identified cutoff value). For this analysis, the same explanatory variables for formal analysis were used except for age. The significance level was set at 5%, and the odds ratios (ORs) and 95% CIs were calculated for each explanatory variable (part 7 in Figure 2).

To estimate the risk of a combination of multiple factors, a decision tree analysis was performed using the significant risk factors in the multivariate logistic regression analysis (part 8 in Figure 2). Chi-Squared Automatic Interaction Detection (CHAID), which divides features using a χ² test, was also used [10]. CHAID is an algorithm that creates multiple cross-tabulation tables between the objective variable and each explanatory variable, and then branches the tree by selecting the variable that is most significant from the χ² test. The analysis conditions were based on a previous study using the same method [11].

Python (version 3.8.8; Anaconda Inc) was used for data sampling, index search, threshold calculation, and multivariate logistic regression analysis, and SPSS Statistics (version 28.0.1.0; IBM Corp) was used for decision tree analysis.

This study was a self-administered, anonymous questionnaire survey (web-based questionnaire) and all data obtained in this study were anonymized and deidentified. Only those who gave their consent to the questionnaire survey and the provision of PHR owned by the harmo Co, Ltd were asked to complete the questionnaire. This study’s protocol was approved by the ethics committee of the Keio University Faculty of Pharmacy (221111-1). No incentives were distributed for survey respondents.

Out of 49,505 patients, 5528 (11.2%) patients opened the survey request notification, 1976 (4%) patients accessed the questionnaire page, and 1501 (3%) patients responded to the questionnaire. Among them, 212 (14.1%) patients answered, “There are tablets or capsules difficult to swallow due to their large size.” Regarding dysphagia, 189 (12.6%) patients reported that they “usually found it difficult to swallow while eating” (Table 1).

Figure 4 shows the ROC of 147 tablets or capsules that were difficult to swallow because of their large size and 624 control tablets or capsules and their AUCs. The highest AUC was obtained for “long diameter + short diameter + thickness” (0.9125). Table 2 presents the threshold values calculated from the illustrated ROCs. The threshold value was calculated to be 21.5 mm for “long diameter + short diameter + thickness,” which had the highest AUC in the index search.

Of the 1501 patients, 86 patients with missing data were excluded. The analysis included 1415 patients (132 patients with tablets or capsules swallowing difficulties owing to the size of the formulations and 1283 patients without swallowing difficulties). As shown in Table S1 in Multimedia Appendix 1, “number of oral medications per day (including powdered medicine and liquid),” “number of tablets or capsules per day,” “number of tablets or capsules per timing,” and “total size of tablets or capsules per day” showed strong correlations. We selected “number of tablets or capsules per timing” as a variable because of its lowest correlation ratio with “taking large tablets or capsules” (Table S2 in Multimedia Appendix 1) and the clinical perspective that it can be easily calculated by health care workers.

The nonbinary variables “age” and “number of tablets or capsules per timing” were classified as binary variables by setting cutoff values (aged 50 years and number of tablets or capsules per timing: 4) based on the results of stratified group comparison (Tables S3 and S4 in Multimedia Appendix 1).

An elevated percentage of patients who showed difficulties swallowing tablets or capsules because of their size were more likely to be female or aged younger than 50 years (Table S5 in Multimedia Appendix 1). The results of the multivariate logistic regression analysis are presented in Table 3. The variables identified as risk factors were aged <50 years (OR 1.59, 95% CI 1.03-2.44), female (OR 2.54, 95% CI 1.70-3.78), dysphagia (OR 3.54, 95% CI 2.22-5.65), and taking large tablets or capsules (OR 9.74, 95% CI 5.19-18.29). Similar trends were obtained from the sensitivity analysis when patients were divided based on whether they were “younger than 50 years” or “aged 50 years or older” (Tables S6 and S7 in Multimedia Appendix 1).

In the subsequent decision tree analysis based on these risk factors, the risk was higher in patients with “taking large tablets or capsules” and “dysphagia” (percentage of patients who reported difficulty in swallowing: 32 out of 94 patients, 34% patients). Additionally, even in the absence of dysphagia, the risk was higher in patients “taking large tablets or capsules” and who were “female” (percentage of patients who reported difficulty in swallowing: 62 out of 299 patients, 20.7% patients; Figure 5).

In this study, we conducted an analysis using a questionnaire with an electronic medication notebook, PHR, and tablets or capsules formulation information.

We identified the size index and threshold of tablets or capsules that patients find difficult to swallow, as well as the risk factors for patients who find tablets or capsules “too large to swallow.” The threshold for the best index was “long diameter + short diameter + thickness” = 21.5 mm, and the risk factors were aged <50 years, female sex, dysphagia, and the use of large tablets or capsules. In addition, patients with “taking large tablets or capsules” and “having dysphagia” and those with “taking large tablets or capsules” and “no dysphagia” and “female” were at relatively high risk.

For the index search and threshold calculation, the PHR-based study design defined “difficult-to-swallow” as well as “not difficult-to-swallow” tablets or capsules. Compared with a previous study based on a geometric analysis of histograms [5], this study enabled conducting a threshold or indicator search using the ROC-AUC and obtaining more generalized findings while strengthening the validity of the previous report.

Although it is difficult to estimate the percentage of people who find tablets or capsules difficult to swallow because of the differences in methods and study populations, it has been shown that 10%-40% of adult patients in general practice find tablets or capsules difficult to swallow [12-15]. Among the patients evaluated in this study, 212 (14.1%) patients felt that some tablets or capsules were difficult to swallow because their sizes fell within the range of values obtained in these studies. A 2007 US survey on dysphagia in general practice patients aged 18 years showed a prevalence of dysphagia of 22.6% [16]. In this study, 189 (12.6%) patients reported that they usually found it difficult to swallow when eating showed a lower percentage than that in the previous report. This may be due to the lower response rate among older adults in this study (who generally have an elevated prevalence of dysphagia) and racial differences than those in previous studies.

The AUC (0.9125) for “long diameter + short diameter + thickness” was extremely good in the index search and threshold calculation, suggesting that this index has high discrimination power as an index of medical tablets or capsule size. Furthermore, the threshold value of 21.5 mm for “long diameter + short diameter + thickness” was similar to that in a previous report [4] (22 mm: a small clinical study of 40 patients) and was considered to have a certain validity.

In multivariate logistic regression analysis, aged <50 years, which was identified as a risk factor, was at odds with the general perception that dysphagia due to tablets or capsules was more frequent in the older age group. However, the result was consistent with those of previous studies, in which younger patients were more likely to report difficulty in swallowing tablets or capsules than older patients [12,17].

These results suggest that the measures commonly taken to make medication easier for older patients may not be considered for younger patients. It is also possible that this was due to younger and middle-aged patients’ unfamiliarity with taking tablets or capsules compared to older patients who take medications daily and are accustomed to taking tablets or capsules. However, we believe that further verification of whether age is a true risk factor for swallowing difficulties is needed. This is because age was not extracted in the DT model, that is, its association with swallowing difficulties may be relatively weaker than that of other variables. In addition, as described above, age was likely influenced by external factors such as the absence of measures to make it easier to swallow tablets and capsules.

Female gender has been identified as a risk factor in several studies [12,13,18]. Schiele et al [12] attributed this to physiological processes and anatomical differences in the dimensions and function of the mouth, pharynx, upper esophageal sphincter, and esophagus, as well as a higher incidence of depression and anxiety disorders. Dysphagia was identified as a risk factor in a previous study in which patients attending general practice were asked to complete a questionnaire, and the same trend was observed [12]. The same trend was observed for large tablets or capsules because the size of the solid dosage form has been cited as a cause of dysphagia in many studies [1-5]. Furthermore, the high OR (9.74) and 95% CI (5.19-18.29) suggest that the size of the medical tablet or capsule is the most important factor in avoiding risk among the risk factors identified in this study. The validity of the threshold value of 21.5 mm for “long diameter + short diameter + thickness,” which was calculated only for patients who reported having a difficult-to-swallow medical tablet or capsule, was strengthened by conducting a factor analysis that included patients who reported having no difficult-to-swallow tablets or capsules.

This study investigated the risk estimation of swallowing by combining factors in a decision tree analysis. In the CHAID model used in this study, the order of the branches reflects the importance of each feature. Since the explanatory variable in the highest branch was “taking large tablets or capsules,” it was the explanatory variable that contributed the most to the objective variable. The risk was higher in patients with “taking large tablets or capsules” and “dysphagia” (32/94, 34% of patients). Additionally, even in the absence of dysphagia, the risk was higher in patients “taking large tablets or capsules” and who were “female” (62/299, 20.7%). In contrast, the risk was low in patients who did not take large tablets or capsules (12/639, 1.9%). These results identified a more detailed group of patients for whom health care workers should provide easier-taking measures and advice when prescribing medications compared with previous studies.

Tablets or capsules that surpass the determined threshold should be subject to formulation adjustments, such as transformation into orally disintegrating tablets, divisible through a dividing line, or formulation of smaller dose units. It is important to design a medical tablet or capsule formulation that does not exceed the threshold size to reduce swallowing difficulty. Furthermore, it is important to design prescriptions and provide guidance to younger patients aged younger than 50 years to ease intake of such formulations, ultimately enhancing compliance with prescribed doses. Moreover, the implementation of straightforward monitoring practices, such as asking patients about their daily experiences with swallowing difficulties during meals and beverages, can serve as an effective means of identifying individuals at a higher risk of dysphagia. Early intervention, including a potential change in dosage form, can then be promptly initiated in response to these findings.

This study acknowledges limitations stemming from the characteristics of the electronic medication notebook and the questionnaire design. Among the limitations associated with the electronic medication notebook, population bias is a notable concern. Furthermore, in this questionnaire survey conducted using harmo, there was a tendency for younger to middle-aged people rather than older people, which could be attributed to the penetration rate. The skewed distribution contrasts with the age profile of hospital outpatients in Japan [19], raising the possibility that the results may have been influenced by this sampling bias. Furthermore, our focus was on those individuals among older people who can complete the questionnaire using their smartphones. A survey conducted among residents of a nursing home revealed that 68% of older participants experienced difficulty in swallowing at mealtime [20]. Therefore, the older people who find it difficult to swallow tablets or capsules may be underestimated. In addition, the medication intake in patients cannot be ensured in that patients actually take the medication, as recorded in the dispensing history, and there is no guarantee that all tablets or capsules taken by patients are recorded in the electronic medication notebook. Furthermore, there is no certainty that all prescribed medications are recorded in the electronic medication book as part of the dispensing history.

The limitations stemming from the questionnaire-based design include the potential for response bias, as this study relies on active participation and interest in swallowing tablets or capsules. Another factor was the high exclusion rate due to missing data, resulting in a reduced sample size from 212 to 132 patients in the factor analysis. Furthermore, we could not use objective measures of dysphagia (such as the 10-item Eating Assessment Tool) due to the need to minimize the number of included questions.

Despite these limitations, this study presents an innovative approach aimed at generating tablets or capsules and prescription considerations using the data collected from electronic medication notebooks. The novelty lies in risk estimation through the combination of various factors and the validation of the usefulness of the PHR infrastructure. We anticipate that the use of PHR will continue to grow in the future, and this study serves as a pioneering step in this promising direction.

Combining the information on patient preferences regarding tablets or capsules obtained from the questionnaire survey of the electronic medication notebook, the user’s PHR and medication formulation information revealed that the most suitable indicator for determining the size of medical tablets or capsules causing swallowing difficulties is “long diameter + short diameter + thickness,” with a threshold of 21.5 mm. In addition, factors such as “aged <50 years,” “female,” “dysphagia,” and “taking large tablets or capsules” were identified as patient backgrounds and prescribing backgrounds that are likely to cause patients to feel difficulty in swallowing large tablets or capsules. Furthermore, patients who ingested “large tablets or capsules” and had “dysphagia” were shown to be at an elevated risk of difficulty in swallowing. Although some sampling biases could not be avoided due to the low response rate (1501/49,505, 3% of patients) and specific survey population (ie, only smartphone users), these findings have significant implications for optimizing the design of patient-friendly formulations and prescriptions.